Questionable package name validation #11800
Comments
From @cpansproutThis is apparently a valid package name: $ perl -le 'q"_#@*$!@*^("->can(can); print -ok' But this is not: $ perl -le 'q"#@*$!@*^(_"->can(can); print -ok' That also means that main::foo is not always equivalent to ::foo: $ perl -le 'main::foo->can(can); print -ok' If we could eliminate that check, it would allow for far more possibilities with UNIVERSAL methods--a bit like autobox. I think this is a pity: $ ./perl -Ilib -le' print foo->CORE::uc' Also, the Unicode Bug affects class method calls. Fixing the above (which would fix this too) is easier than fixing just this: $ ./perl -Ilib -Mutf8 -le '$p = "þackage"; $p->can(can)' $ ./perl -Ilib -Mutf8 -le '$p = "þackage"; utf8::downgrade $p; $p->can(can)' It doesn’t even work when the package exists. And to drive the point home: $ perl -e 'sub foo { warn $::Moose::VERSION } use Moose' Flags: Site configuration information for perl 5.15.4: Configured by sprout at Wed Nov 2 09:06:14 PDT 2011. Summary of my perl5 (revision 5 version 15 subversion 4) configuration: Locally applied patches: @INC for perl 5.15.4: Environment for perl 5.15.4: |
From @ikegamiOn Sun, Dec 11, 2011 at 4:35 PM, Father Chrysostomos <
And feature unicode_strings doesn't help. |
|
The RT System itself - Status changed from 'new' to 'open' |
From @HugmeirOn Tue, Dec 13, 2011 at 5:48 PM, Eric Brine <ikegami@adaelis.com> wrote:
Whoops, that was me. Fixing this is one line in pp_hot.c (patch attached), A valid class name, as far as the left side of -> is concerned, is either The IO part is the real kicker though: You can make *anything* a valid |
From @Hugmeir0001-pp_hot.c-First-letter-of-latin-1-classnames-wasn-t.patchFrom a9ad2aee8acc54e97fdd09e75ae9390b5b2807b2 Mon Sep 17 00:00:00 2001
From: Brian Fraser <fraserbn@gmail.com>
Date: Tue, 13 Dec 2011 19:32:48 -0300
Subject: [PATCH] pp_hot.c: First letter of latin-1 classnames wasn't being checked correctly.
Previously the first letter for latin-1 classnames was being mischecked, only
allowing ASCII, which caused an instance of the Unicode Bug for downgradable
classnames.
---
pp_hot.c | 2 +-
t/uni/package.t | 19 ++++++++++++++++++-
2 files changed, 19 insertions(+), 2 deletions(-)
diff --git a/pp_hot.c b/pp_hot.c
index 70eb5a1..a2d6f91 100644
--- a/pp_hot.c
+++ b/pp_hot.c
@@ -2973,7 +2973,7 @@ S_method_common(pTHX_ SV* meth, U32* hashp)
if (!packname ||
((UTF8_IS_START(*packname) && DO_UTF8(sv))
? !isIDFIRST_utf8((U8*)packname)
- : !isIDFIRST(*packname)
+ : !isIDFIRST_L1((U8)*packname)
))
{
Perl_croak(aTHX_ "Can't call method \"%"SVf"\" %s",
diff --git a/t/uni/package.t b/t/uni/package.t
index 317ddd4..bb9092b 100644
--- a/t/uni/package.t
+++ b/t/uni/package.t
@@ -6,7 +6,7 @@ BEGIN {
require './test.pl';
}
-plan (tests => 16);
+plan (tests => 18);
use utf8;
use open qw( :utf8 :std );
@@ -92,3 +92,20 @@ ok 1, "sanity check. If we got this far, UTF-8 in package names is legal.";
eval q[package ��� {];
like $@, qr/\AMissing right curly /, "comp/package_block.t test";
}
+
+# perl #105922
+
+{
+ my $latin_1 = "��ackage";
+ my $utf8 = "��ackage";
+ utf8::downgrade($latin_1);
+ utf8::upgrade($utf8);
+
+ local $@;
+ eval { $latin_1->can("yadda") };
+ ok(!$@, "latin1->meth works");
+
+ local $@;
+ eval { $utf8->can("yadda") };
+ ok(!$@, "utf8->meth works");
+}
--
1.5.6.3
|
From tchrist@perl.com
Oh darn it. I had convinced myself that Perl identifiers could And which ones can start with \p{Nd} is also anything but clear. Not clear, not clean. --tom |
From @HugmeirOn Tue, Dec 13, 2011 at 8:12 PM, Tom Christiansen <tchrist@perl.com> wrote:
That's rightish too. But do note that I said class names on the left side |
From @cpansproutOn Tue Dec 13 14:58:38 2011, Hugmeir wrote:
That’s not a class method call. That’s an IO method. $ perl -E 'open q<#@*$!@*^(_>, ">", undef; q<#@*$!@*^(_>->oeunt' And 5.9.5 had a bug making constants turn into IOs: perl5.9.5 -Mconstant=Just,0 -le 'sub :-) -- Father Chrysostomos |
From @cpansproutOn Tue Dec 13 14:58:38 2011, Hugmeir wrote:
Thank you. Applied as d47f310. -- Father Chrysostomos |
|
@cpansprout - Status changed from 'open' to 'resolved' |
|
@cpansprout - Status changed from 'resolved' to 'open' |
From @doySo is the way to move forward here to just make any string valid on the -doy |
From @cpansproutOn Tue Jul 03 14:04:57 2012, doy wrote:
Yes, I think so. It’s on my to-do list, but my to-do list is *long*. :-) In other words, I would be happy for someone to get to this before I do. -- Father Chrysostomos |
From [Unknown Contact. See original ticket]On Tue Jul 03 14:04:57 2012, doy wrote:
Yes, I think so. It’s on my to-do list, but my to-do list is *long*. :-) In other words, I would be happy for someone to get to this before I do. -- Father Chrysostomos |
From @rjbsOn Tue Jul 03 14:04:57 2012, doy wrote:
Yes. |
From @cpansproutOn Wed Jul 04 05:46:31 2012, rjbs wrote:
Test::More fails its tests with the attachment, because it’s doing this: elsif( $error =~ /Can't call method "isa" without a package/ ) { but "anything"->isa("Wibble") will simply return false now, instead of That means Test::More::isa_ok may output: # Failed test 'My Wibble isa Wibble' instead of what its tests are checking for: # Failed test 'My Wibble isa Wibble' (That ‘it's a ''’ is another bug, reported as #78204.) I think Test::More’s tests will need a little bit of adjustment. -- Father Chrysostomos |
From @cpansproutFrom 8eff703c953b1daece21d3ef7bda95618fac1a0a Mon Sep 17 00:00:00 2001 The rules for filtering out what do not look like package names are This commit simply lets any non-empty string be used as a package I made an exception for the empty string because it messes up caches Inline Patchdiff --git a/pp_hot.c b/pp_hot.c
index 77b707c..6366aee 100644
--- a/pp_hot.c
+++ b/pp_hot.c
@@ -2960,12 +2960,14 @@ S_method_common(pTHX_ SV* meth, U32* hashp)
PERL_ARGS_ASSERT_METHOD_COMMON;
if (!sv)
+ undefined:
Perl_croak(aTHX_ "Can't call method \"%"SVf"\" on an undefined value",
SVfARG(meth));
SvGETMAGIC(sv);
if (SvROK(sv))
ob = MUTABLE_SV(SvRV(sv));
+ else if (!SvOK(sv)) goto undefined;
else {
GV* iogv;
STRLEN packlen;
@@ -2994,17 +2996,11 @@ S_method_common(pTHX_ SV* meth, U32* hashp)
!(ob=MUTABLE_SV(GvIO(iogv))))
{
/* this isn't the name of a filehandle either */
- if (!packname ||
- ((UTF8_IS_START(*packname) && DO_UTF8(sv))
- ? !isIDFIRST_utf8((U8*)packname)
- : !isIDFIRST_L1((U8)*packname)
- ))
+ if (!packname || !packlen)
{
- /* diag_listed_as: Can't call method "%s" without a package or object reference */
- Perl_croak(aTHX_ "Can't call method \"%"SVf"\" %s",
- SVfARG(meth),
- SvOK(sv) ? "without a package or object reference"
- : "on an undefined value");
+ Perl_croak(aTHX_ "Can't call method \"%"SVf"\" "
+ "without a package or object reference",
+ SVfARG(meth));
}
/* assume it's a package name */
stash = gv_stashpvn(packname, packlen, packname_is_utf8 ? SVf_UTF8 : 0);
diff --git a/t/op/method.t b/t/op/method.t
index 09f6ee3..8f6bfb8 100644
--- a/t/op/method.t
+++ b/t/op/method.t
@@ -13,7 +13,7 @@ BEGIN {
use strict;
no warnings 'once';
-plan(tests => 98);
+plan(tests => 103);
@A::ISA = 'B';
@B::ISA = 'C';
@@ -417,3 +417,15 @@ eval { () = undef; new {} };
like $@,
qr/^Can't call method "new" without a package or object reference/,
'Err msg from new{} when stack contains undef';
+
+sub flomp { "flimp" }
+sub main::::flomp { "flump" }
+is "::"->flomp, 'flump', 'method call on ::';
+is "::main"->flomp, 'flimp', 'method call on ::main';
+eval { ""->flomp };
+like $@,
+ qr/^Can't call method "flomp" without a package or object reference/,
+ 'method call on empty string';
+is "3foo"->CORE::uc, '3FOO', '"3foo"->CORE::uc';
+{ no strict; @{"3foo::ISA"} = "CORE"; }
+is "3foo"->uc, '3FOO', '"3foo"->uc (autobox style!)';
diff --git a/t/run/fresh_perl.t b/t/run/fresh_perl.t
index cd5899a..376ceaf 100644
--- a/t/run/fresh_perl.t
+++ b/t/run/fresh_perl.t
@@ -81,7 +81,7 @@ $array[128]=1
########
$x=0x0eabcd; print $x->ref;
EXPECT
-Can't call method "ref" without a package or object reference at - line 1.
+Can't locate object method "ref" via package "961485" (perhaps you forgot to load "961485"?) at - line 1.
########
chop ($str .= <DATA>);
######## |
From [Unknown Contact. See original ticket]On Wed Jul 04 05:46:31 2012, rjbs wrote:
Test::More fails its tests with the attachment, because it’s doing this: elsif( $error =~ /Can't call method "isa" without a package/ ) { but "anything"->isa("Wibble") will simply return false now, instead of That means Test::More::isa_ok may output: # Failed test 'My Wibble isa Wibble' instead of what its tests are checking for: # Failed test 'My Wibble isa Wibble' (That ‘it's a ''’ is another bug, reported as #78204.) I think Test::More’s tests will need a little bit of adjustment. -- Father Chrysostomos |
From @cpansproutOn Wed Jul 04 12:36:34 2012, sprout wrote:
which is also on the sprout/package-names-113974 branch. -- Father Chrysostomos |
From [Unknown Contact. See original ticket]On Wed Jul 04 12:36:34 2012, sprout wrote:
which is also on the sprout/package-names-113974 branch. -- Father Chrysostomos |
From @rjbs* Father Chrysostomos via RT <perlbug-comment@perl.org> [2012-07-04T15:36:35]
I think so, too. Schwern, what say you? -- |
From @schwern-----BEGIN PGP SIGNED MESSAGE----- On 2012.7.5 7:53 PM, Ricardo Signes wrote:
I have no objection. - -- iEYEARECAAYFAk/2WrgACgkQWMohlhD1QyfosgCfQRSIoGkdAK7JGRJre4eKalZY |
From @cpansproutFixed in 7156e69. Father Chrysostomos |
From [Unknown Contact. See original ticket]Fixed in 7156e69. Father Chrysostomos |
|
@cpansprout - Status changed from 'open' to 'resolved' |
Migrated from rt.perl.org#105922 (status was 'resolved')
Searchable as RT105922$
The text was updated successfully, but these errors were encountered: